Beruflich Dokumente
Kultur Dokumente
Bitcoin’s Growing digital cash. As with any digital informa- a topic of heavy debate over the past
tion, a digital token may be reproduced few years. It is a process that makes
Energy Problem relatively easily. If this were to happen Bitcoin extremely energy-hungry by
Alex de Vries1,* in Bitcoin it would lead to inflation in design, as the currency requires a
the digital currency and devalue it rela- huge amount of hash calculations
tive to other currencies. In turn, this for its ultimate goal of processing finan-
would compromise user trust in the cur- cial transactions without intermediaries
rency.1 Nakamoto’s solution involved (peer-to-peer). We cannot observe this
‘‘timestamping transactions by hashing hashrate directly, but it is possible to
them into an ongoing chain of hash- derive this number from the observable
based proof-of-work.’’ The proof-of- difficulty and the actual time required
work was specifically said to involve to mine new blocks for the blockchain.
‘‘scanning for a value that when hashed, As per mid-March 2018, about 26 quin-
such as with SHA-256, the hash begins tillion hashing operations are per-
with a number of zero bits.’’2 The num- formed every second and non-stop by
ber of attempts to find such a hash, the Bitcoin network (Figure 1). At the
made every second, is what we call same time, the Bitcoin network is only
the hashrate. Once a node finds a processing 2–3 transactions per second
hash that satisfies the required number (around 200,000 transactions per day).
of zero bits, it transmits the block it was This means that the ratio of hash
working on to the rest of the network. calculations to processed transactions
The other nodes in the network then ex- is 8.7 quintillion to 1 at best. The pri-
press their acceptance by starting to mary fuel for each of these calculations
The electricity that is expended in the
create the next block for the blockchain is electricity.
process of mining Bitcoin has become
using the hash of the accepted block.
a topic of heavy debate over the past
The finder of the block is rewarded for Network Power Usage
few years. It is a process that makes
the efforts with a special transaction. Trying to measure the electricity
Bitcoin extremely energy-hungry by
Creators of a block are currently al- consumed by the Bitcoin mining ma-
design, as the currency requires a
lowed to send 12.5 newly created coins chines producing all those hash calcula-
huge amount of hash calculations for
to an address of their choosing. This tions remains a challenge to date. Even
its ultimate goal of processing financial
is a fixed reward that halves every though we can easily estimate the total
transactions without intermediaries
four years (210,000 blocks). On top of computational power of the Bitcoin
(peer-to-peer). The primary fuel for
the fixed reward a variable amount of network, it provides only little informa-
each of these calculations is electricity.
transaction fees is received as well. tion on the underlying machines and
The Bitcoin network can be estimated
The reward provides an incentive to their respective power use. A hashrate
to consume at least 2.55 gigawatts of
participate in this type of network. The of 14 terahashes per second can either
electricity currently, and potentially
more computational power one has, come from a single Antminer S9 running
7.67 gigawatts in the future, making
the bigger the share of all distributed on just 1,372 W, or more than half a
it comparable with countries such as
rewards that go to that miner. To keep million Playstation-3 devices running
Ireland (3.1 gigawatts) and Austria
the flow of rewards stable, the network on 40 MW (as a single Playstation-3 de-
(8.2 gigawatts). Economic models tell
self-adjusts the difficulty of hash calcu- vice has a hashrate of 21 megahashes
us that Bitcoin’s electricity consumption
lations, so new blocks are only created per second and a power use of 60 W).
will gravitate toward the latter number.
once every 10 min on average. Naka- It is also not possible to observe the
A look at Bitcoin miner production esti-
moto compared the creation of new exact number of connected devices.
mates suggests that this number could
coins in this way with gold mining The Bitcoin network is estimated to
already be reached in 2018.
(hence the term Bitcoin mining), and have around 10,000 connected nodes,
noted that ‘‘in our case, it’s CPU time but a single node in the network can
Introduction and electricity that is expended.’’ represent either one or many machines.
When Bitcoin was introduced in 2008,
Satoshi Nakamoto presented a solution The electricity that is expended in the Still, estimating the power consump-
for the double-spending problem in process of mining Bitcoin has become tion of the Bitcoin network using the
thus expect a total electricity con- and 27 machines per wafer, and given devices) was shut down after ‘‘not
sumption of 7.67 GW. that these production rates are main- paying for several million kilowatt-
tained throughout the year, Bitmain hours of electricity.’’14
Thanks to its simplicity, the aforemen- could produce up to 6.5 million
tioned approach became the founda- Antminer S9 machines in 2018. These Less malicious reasons for an agent
tion of the Bitcoin Energy Consumption machines would have a combined elec- to mine Bitcoin at a loss might include
Index,10 but knowing where Bitcoin’s tricity consumption of 8.92 GW. This motivations such as being able to
electricity consumption is heading exceeds the expected electricity con- obtain Bitcoin completely anony-
does not provide us a with final esti- sumption of 7.67 GW from before, mously, libertarian ideology (support-
mate for the network’s current con- which therefore seems to be within ing a payment network that does not
sumption. It is important to note that Bitmain’s production potential for rely on a central authority), or specula-
the index assumes it may currently 2018. It is worth noting that these ma- tive reasons. None of these situations
take around a year before the expected chines might not all be finalized and would be properly captured by any of
electricity consumption is actually delivered in 2018, but at the same time the discussed methods.
reached. Especially after strong price Bitmain, with a claimed market share
increases, one needs to allow for a suf- of 70%,8 is not the only contributor to Conclusion
ficient amount of time for the produc- the industry’s total production potential This paper has outlined various
tion of new hardware. this year. methods that are currently used in
determining the current and future elec-
Bitcoin Miner Production The aforementioned also marks the tricity consumption of the Bitcoin
Bitcoin miner manufacturers tend to be first time that Bitcoin miner production network. These methods tell us that
very secretive about their production has been estimated with the help of the Bitcoin network consumes at least
output, but Morgan Stanley managed upstream (chip) production numbers. 2.55 GW of electricity currently, and
to work around this problem by looking Given the ongoing secrecy of Bitcoin that it could reach a consumption of
at the TSMC instead. TSMC supplies miner manufacturers, this could prove 7.67 GW in the future, making it compa-
the chips for Bitmain, making it possible to be a valuable addition to the toolkit rable with countries such as Ireland
to come up with a chip-based produc- for substantiating trends in Bitcoin’s (3.1 GW) and Austria (8.2 GW).15 Addi-
tion potential. Specifically, Morgan electricity consumption. tionally, economic models tell us that
Stanley estimated that TSMC had or- Bitcoin’s electricity consumption will
ders from Bitmain ‘‘for 15–20k wafer- Limitations gravitate toward the latter figure. A
starts per month’’ for the first quarter Lastly, it is important to keep in mind look at Bitcoin miner production esti-
of 2018. This figure was independently that all of the methods discussed as- mates suggests that this figure could
confirmed by Ark Invest analyst James sume rational agents. There may be already be reached in 2018. With the
Wang, who wrote that ‘‘Bitmain is various reasons for an agent to mine Bitcoin network processing just
buying 20k 16 nm wafers a month’’ even when this isn’t profitable, and in 200,000 transactions per day, this
specifically (used for building the Ant- some cases costs may not play a role means that the average electricity
miner S9 and T9).11 at all when machines and/or electricity consumed per transaction equals at
are stolen or abused. In one case a least 300 kWh, and could exceed
With each 16-nm wafer capable of sup- researcher misused National Science 900 kWh per transaction by the end of
plying chips for about ‘‘27–30 Bitcoin Foundation-funded supercomputers to 2018. The Bitcoin development com-
mining rigs’’, Bitmain could produce mine $8,000–$10,000 worth of Bitcoin. munity is experimenting with solutions
around half a million of its most efficient The operation ended up costing the such as the Lightning Network to
Bitcoin mining machines per month.12 university $150,000.13 More recently, improve the throughput of the network,
Assuming 20,000 wafers per month a mining facility in Russia (with 6,000 which may alleviate the situation. For